MLS-C01 Question 274
Select 3You are tasked with deploying a custom machine learning model using Amazon SageMaker. To ensure that your model runs in a containerized environment with the necessary dependencies, you decide to create a custom Docker image. Which steps are required to successfully build and use the Docker container for SageMaker?
- A
Create a Dockerfile that defines the base image, dependencies, and entry point for your model.
- B
Push the Docker image to an Amazon S3 bucket for SageMaker to pull during deployment.
- C
Tag the Docker image with the ECR repository URI before pushing it to Amazon Elastic Container Registry (ECR).
- D
Use the Amazon SageMaker SDK to specify the custom Docker image URI when creating the model or endpoint.
- E
Install the SageMaker Python SDK inside the Docker container.
Show answer and explanation
Correct answers: A, C, D
Explanation
To use a custom Docker container in Amazon SageMaker, you must define the container using a Dockerfile, build and tag the image with the Amazon ECR repository URI, and push it to ECR. The SageMaker SDK or console can then reference the custom container URI for deployment. SageMaker does not interact with Docker images stored in Amazon S3, and installing the SageMaker Python SDK in the container is unnecessary unless explicitly needed for custom logic.
- A. Correct.
Correct. A Dockerfile is essential for specifying the base image, dependencies, and commands needed to set up your custom container for SageMaker.
- B. Incorrect.
Incorrect. Amazon SageMaker pulls Docker images from Amazon Elastic Container Registry (ECR), not from Amazon S3.
- C. Correct.
Correct. Tagging the Docker image with the ECR repository URI is required before pushing it to ECR, so SageMaker can locate and pull it.
- D. Correct.
Correct. When deploying the model or endpoint, you need to provide the custom Docker image URI in the SageMaker SDK or console.
- E. Incorrect.
Incorrect. While the SageMaker Python SDK is used to interact programmatically with SageMaker, it is not required inside the Docker container for the model to run.